Tetsuwan Scientific is building AI-powered robotic scientists that can conduct experiments

San Francisco-based startup Tetsuwan Scientific is building artificial intelligence (AI) robotics that can perform the tasks of a scientist. Co-founders, CEO Cristian Pons and Chief Technology Officer (CTO) Theo Schaefer brought the startup out of the closet in November after a successful seed round of funding. The company’s goal is to create intelligent software that can be integrated with laboratory robotics to automate the entire process of scientific discovery and invention, from formulating a hypothesis to running experiments and drawing conclusions.

AI-powered robotics scientists create

Founded in 2023, the startup had been working in secret for the past year and a half to create its first product – an AI scientist who can run experiments. It is now out of stealth and is currently working with La Jolla Labs in RNA therapeutic drug development. On its website, the startup has detailed its vision and the first product it is working on. Notably, it does not have any products in the public domain yet.

Highlighting the problem statement it aims to solve, the startup says automation in science focuses on high volume of experiments rather than high variety. This is because laboratory robots currently require extensive programming to replicate specific protocols. However, it has led to the creation of a system that builds assembly lines instead of robots that could be helpful to scientists, the company said.

Tetsuwan Scientific said the problem is that robots cannot understand scientific intent, and thus, cannot perform an experiment on their own. However, looking at generic AI models, the company says, it is now possible to bridge this communication gap and teach robots to act like a scientist. This is a two-pronged problem that requires intelligent software combined with versatile robotics hardware.

In an interview with TechCrunch, Pons highlighted that large language models (LLMs) can bridge the software gap by allowing developers to communicate scientific intent to robots without needing to write thousands of lines of code. The CEO highlighted that the Retrieval-Augmented Generation (RAG) framework can also help keep AI hallucinations down.

According to the publication, Tetsuwan Scientific is building non-humanoid robots. These robots, which are also showcased on the website, are large square-shaped glass-like structures that are said to evaluate results and make changes to scientific experiments without the need for human intervention. These robots are said to be powered by AI software and sensors to gain knowledge about technical parameters such as calibration, fluid class characterization and other properties.

Specifically, the startup is currently in the early stages towards its ultimate goal of building independent robotic AI scientists that can automate the entire scientific process and invent things.